# 582,000 AI Projects, One Slow Language: Why Python Won

> C is far faster, yet nearly half of new AI projects on GitHub use Python. The answer is what actually runs the math.

*2026-11-02T13:42:20Z*

Why is Python used for AI despite being slower than C? GitHub reports that Python powers nearly half of all new AI repositories. That looks like a contradiction until you see what actually runs the math.

## In 60 seconds (the video)

1. **582,000 AI projects use a slow language.** Why?
2. **C is much faster.** Yet nearly half of new AI projects on GitHub are in Python.
3. **Christmas 1989.** Guido van Rossum had time on his hands and started a language that is easy to read.
4. **Easy has a price.** Hand it millions of numbers and a Python loop crawls.
5. **Python stopped running.** A library like NumPy takes the order and runs it in C.
6. **It didn't win on speed.** It won by getting the faster one to run for it.

## What the 582,000 means

GitHub's Octoverse 2025 report, covering September 1, 2024 to August 31, 2025, says: "Python now powers nearly half of all new AI repositories (582,196; +50.7% YoY)."

Three things to keep in mind:

- These are repositories on GitHub, and GitHub's own wording is that Python powers them. The report assigns each repository one primary language, so the count most likely means repositories whose primary language is Python.
- GitHub's metrics reflect public activity unless noted otherwise.
- The report doesn't give the total behind "nearly half", so don't divide 582,196 by the 4.3 million AI-related repositories it mentions elsewhere.

## It started in a holiday week

In his foreword to *Programming Python*, Guido van Rossum says that in December 1989 he was looking for a hobby programming project for the week around Christmas, while his office at a research lab in Amsterdam was closed. The Python FAQ quotes him: "During the 1989 Christmas holidays, I had a lot of time on my hand, so I decided to give it a try."

There was a work reason too. His group on the Amoeba operating system needed a better way to do system administration than C programs or Bourne shell scripts.

The name has nothing to do with snakes. The Python tutorial says the language is "named after the BBC show 'Monty Python's Flying Circus' and has nothing to do with reptiles."

Readability stayed central. Python groups statements by indentation instead of brackets, and in 1999 van Rossum described it as "easy to learn, read, and use".

## Why loops are slow, and how NumPy gets around it

This is the plain Python way to multiply two lists:

```python
c = []
for i in range(len(a)):
    c.append(a[i]*b[i])
```

The NumPy documentation says that if `a` and `b` each contain millions of numbers, "we will pay the price for the inefficiencies of looping in Python." The same task runs much more quickly in C, but you lose the simplicity of Python.

NumPy keeps both:

```python
c = a * b
```

That one line does the same work, and according to the docs the operation is "speedily executed by pre-compiled C code", at near-C speeds. You write the order in Python, and C does the running.

## Learn it if

- You want to get into AI. Python remains GitHub's leading language for AI and data science work.
- You want your computer to do repetitive work. The Python tutorial opens with exactly that: renaming files, search and replace, tasks you would like to automate.

## FAQ

### Is Python the slowest programming language?
The sources here don't claim that. NumPy's documentation compares Python loops with C and says the same task runs much more quickly in C.

### Does all AI code run through NumPy?
No. NumPy is the example documented here. PyTorch, for instance, describes itself as deeply integrated into Python rather than a binding into a monolithic C++ framework.

## Sources

- [GitHub Octoverse 2025](https://github.blog/news-insights/octoverse/octoverse-a-new-developer-joins-github-every-second-as-ai-leads-typescript-to-1/): the 582,196 figure, the report period and the methodology.
- [Python General FAQ](https://docs.python.org/3/faq/general.html): Guido van Rossum's account of how Python started.
- [Foreword to *Programming Python*, Guido van Rossum, May 1996](https://www.python.org/doc/essays/foreword/): December 1989 and the closed office.
- [Python Tutorial, Whetting Your Appetite](https://docs.python.org/3/tutorial/appetite.html): the name, indentation and automation.
- [Computer Programming for Everybody, July 1999](https://www.python.org/doc/essays/cp4e/): "easy to learn, read, and use".
- [What is NumPy?](https://numpy.org/doc/stable/user/whatisnumpy.html): Python loops and pre-compiled C code.
- [PyTorch README](https://github.com/pytorch/pytorch): how PyTorch describes itself.
- Guido van Rossum portrait in the video: Daniel Stroud, [CC BY-SA 4.0](https://commons.wikimedia.org/wiki/File:Guido-portrait-2014-drc.jpg).

## Next steps

- [Top Programming Languages in 2026: Rankings and How to Choose](https://fadymondy.com/en/blog/top-programming-languages)
- [What Is Null in Programming? The Billion-Dollar Mistake](https://fadymondy.com/en/blog/what-is-null)

---

Source: https://fadymondy.com/en/blog/why-slow-python-won-ai